Closed s13n closed 3 years ago
NMOS IS-04 requires that the GMID of the PTP reference clock be given with lowercase hex number notation. The corresponding SDP attribute a=ts-refclk has it the other way around, i.e. uppercase hex only.
https://github.com/AMWA-TV/nmos-discovery-registration/issues/144
Somewhere a conversion must happen either way, but I don't see where.
Furthermore, nmos-cpp seems to assume that the a=ts-refclk attribute is always in the media section of the SDP file. Indeed that is what ST2110 (stupidly) demands.
I'm well aware of the (insert adjective of your choice) additional demands added by ST 2110 beyond RFC 7273. I raised this previously with the ST 2110 team, and was told it was to make things "simpler". 🤷♂
So I think nmos-cpp should at least recognise and properly handle either style when receiving an SDP. At the moment, it insists on ST2110 on input, too.
Please tell if you want me to produce a patch.
Yep, updating nmos::get_session_description_sdp_parameters
to use session-level values for ts-refclk
and mediaclk
as the default, as per RFC 7273 Section 4.8, would be nice.
Gosh, time to update my repository. Sorry to bother you with old stuff.
I raised this previously with the ST 2110 team, and was told it was to make things "simpler".
Ouch. That's not my concept of "simpler".
I will try to produce a patch for session-level ts-refclk and mediaclk shortly.
Yep, updating
nmos::get_session_description_sdp_parameters
to use session-level values forts-refclk
andmediaclk
as the default, as per RFC 7273 Section 4.8, would be nice.
See https://github.com/aholzinger/nmos-cpp/commit/80f41a8679a15f85e6a2cbd75e77600250bdfbcb
NMOS IS-04 requires that the GMID of the PTP reference clock be given with lowercase hex number notation. The corresponding SDP attribute
a=ts-refclk
has it the other way around, i.e. uppercase hex only. Somewhere a conversion must happen either way, but I don't see where.Furthermore, nmos-cpp seems to assume that the
a=ts-refclk
attribute is always in the media section of the SDP file. Indeed that is what ST2110 (stupidly) demands. Other relevant standards don't have this restriction, for example AES67, and indeed the RFC 7273 they all depend on. One would have thought that the normal case is a common reference clock, so that putting it into the global section of the SDP file would be the most prevalent case, if it weren't for ST2110.So I think nmos-cpp should at least recognise and properly handle either style when receiving an SDP. At the moment, it insists on ST2110 on input, too.
Please tell if you want me to produce a patch.